[
https://issues.apache.org/jira/browse/HADOOP-6323?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12788847#action_12788847
]
Aaron Kimball commented on HADOOP-6323:
---------------------------------------
With Writable serialization, yes, you've always had to pre-declare the
datatypes for each of the four key/value types emitted during a job dataflow.
(The input types were implicit from the InputFormat/RecordReader.)
I don't think it's necessary to dynamically determine the schema for emitted
data during jobs in Avro either. I think if it were, someone would have already
asked for some sort of "DelayedWritable" notion to be integrated in MapReduce
already -- which I don't think we've seen. It might be nice to support this
eventually, but let's wait for demand.
> Serialization should provide comparators
> ----------------------------------------
>
> Key: HADOOP-6323
> URL: https://issues.apache.org/jira/browse/HADOOP-6323
> Project: Hadoop Common
> Issue Type: New Feature
> Components: io
> Reporter: Doug Cutting
> Assignee: Aaron Kimball
> Attachments: HADOOP-6323.2.patch, HADOOP-6323.3.patch,
> HADOOP-6323.4.patch, HADOOP-6323.5.patch, HADOOP-6323.6.patch,
> HADOOP-6323.patch
>
>
> The Serialization interface should permit one to create raw comparators.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
